kwame nkrumah university of science and technology kumasi ...
kwame nkrumah university of science and technology college
Transcript of kwame nkrumah university of science and technology college
KWAME NKRUMAH UNIVERSITY OF SCIENCE AND TECHNOLOGY
COLLEGE OF ENGINEERING
FACULTY OF ELECTRICAL AND COMPUTER ENGINEERING
DEPARTMENT OF TELECOMMUNICATION ENGINEERING
PERFORMANCE EVALUATION OF A GSM/GPRS CELLULAR NETWORK USING
THE CSSR WITH DIRECT TCH ASSIGNMENT FEATURE.
by
Aninyie, Paula BSc. (Hons)
May, 2012
PERFORMANCE EVALUATION OF A GSM/GPRS CELLULAR NETWORK USING
THE CSSR WITH DIRECT TCH ASSIGNMENT FEATURE.
by
Aninyie, Paula BSc. (Hons.)
A Thesis submitted to the Department of Telecommunication Engineering,
Kwame Nkrumah University of Science and Technology
in partial fulfillment of the requirements for the Degree
of
MASTER OF SCIENCE
Faculty of Electrical & Computer Engineering
College of Engineering
May, 2012
ii
DECLARATION
I hereby declare that this submission is the result of my own effort towards the MSc.
Telecommunication Engineering, and that to the best of my knowledge, this study contains no
material previously published by another person nor submitted for the award of any other degree
of the Kwame Nkrumah University of Science and Technology, except where acknowledgement
has been duly made in the text.
ANINYIE, PAULA (PG 1396607) ……………………. ……………………
SIGNATURE DATE
CERTIFIED BY:
DR. K. DIAWUO ……………………. ……………………
SUPERVISOR SIGNATURE DATE
CERTIFIED BY:
DR. P.Y. OKYERE ……………………. ……………………
HEAD OF DEPARTMENT SIGNATURE DATE
iii
ABSTRACT
All GSM operators use Key Performance Indicators (KPIs) to judge their network performance
and evaluate the Quality of Service (QoS) regarding end user perceived quality. KPIs are
becoming increasingly important in the context of network rollouts as well as within mature
network optimization cycles. The performance of the mobile network is measured based on
several counters describing the most important events over a measurement period. The KPIs are
derived with the help of these counters using different formulations. Call Setup Success Rate
(CSSR) is one of the most important KPIs used by all mobile operators.
The study made a comparative analysis of measured CSSR and an estimated CSSR which was
evaluated using the CSSR formulation presented in [17]. Results obtained from graphical
representations showed that the measured CSSR were within acceptable levels though a number
of sharp drops were recorded. The estimated CSSR, on the other hand, showed significant
improvements in areas where sharp drops in CSSR values were recorded for the measured
CSSR. Significantly high R square values of close to 1 representing a high predictive ability
from the regression analysis of the estimated CSSR were also recorded. It was concluded that the
implementation of the CSSR formulation be extended to CSSR measurements to ensure
increased subscriber satisfaction.
iv
ACKNOWLEDGEMENT
This thesis and the entire MSc. Telecommunication Engineering Programme has been made
possible through the strength and wisdom bestowed on me by the Almighty God. I am convinced
without Him I could not have reached this stage.
Secondly, I would like to express my profound gratitude to my supervisor Dr. K. Diawuo for his
meticulous guidance which paved the way for this final output.
To my parents Mr. and Mrs. Norbert Aninyie, and my siblings Aaron and Rufina for the
wonderful support and guidance. You have indeed made a difference in my life and I appreciate
that very much.
I owe a great deal of appreciation to Mr. Awuni who gave me sleepless nights through constant
reminders of how I need to finish this thesis. I tell you, I have finished.
I also recognise with gratitude, the support of Mr. Duah of the Department of Electrical
Engineering, KNUST, Kumasi.
I cannot afford to overlook the contribution of my course mate, Koffi Agbeblewu Dotche.
Finally, I wish to acknowledge the diverse support and encouragement from friends and relatives
such as Ing. Kwame Donkor Debrah, Mabel Hoedoafia and Erwin Alhassan.
v
TABLE OF CONTENTS
Title Page ....................................................................................................................................i
Declaration .............................................................................................................................. ii
Abstract .................................................................................................................................. iii
Acknowledgement ................................................................................................................... iv
Table of Contents ..................................................................................................................... v
List of Figures..........................................................................................................................vii
List of Tables ......................................................................................................................... viii
Abbreviations .......................................................................................................................... ix
CHAPTER ONE ....................................................................................................................... 1
GENERAL INTRODUCTION ................................................................................................. 1
1.1 Background of the Study ................................................................................................. 2
1.2 Objectives of the Study .................................................................................................... 2
1.3 Justification of the Study ................................................................................................. 3
1.4 Organisation of the Study ................................................................................................ 3
CHAPTER TWO ...................................................................................................................... 4
LITERATURE REVIEW ......................................................................................................... 4
2.1 Brief History ..................................................................................................................... 4
2.2 GSM Overview ................................................................................................................. 5
2.3 GSM Architecture ............................................................................................................ 6
2.3.1 Mobile Station ............................................................................................................. 6
2.3.2 Base Station System .................................................................................................... 7
2.3.3 Network and Switching Subsystem.............................................................................. 8
2.3.3.1 Home Location Register ....................................................................................... 9
2.3.3.2 Visitor Location Register ..................................................................................... 9
2.3.4 Radio Interface ............................................................................................................ 9
2.4 GPRS Overview .............................................................................................................. 14
2.5 Performance Management ............................................................................................. 16
vi
CHAPTER THREE ................................................................................................................ 30
METHODOLOGY .................................................................................................................. 30
3.1 Research Statement ........................................................................................................ 30
3.2 Data Collection ............................................................................................................... 30
3.3 Sample size and data processing ................................................................................... 30
CHAPTER FOUR ................................................................................................................... 32
RESULTS AND ANALYSIS .................................................................................................. 32
4.1 Presentation of Results ................................................................................................... 34
4.1.1 Graphical Representation of Measured and Estimated CSSR ..................................... 34
4.1.2 Regression Analysis of Estimated CSSR .................................................................... 37
4.2 Interpretation of Results ................................................................................................ 40
4.2.1 Interpretation of Graphical Representations ............................................................... 40
4.2.2 Interpretation of Regression Analysis of Estimated CSSR .......................................... 41
CHAPTER FIVE........................................................................................................ ……….44
CONCLUSION AND RECOMMENDATION .................................................................. …44
REFERENCES ........................................................................................................................ 45
APPENDIX A: TOP LEVEL KPIs ........................................................................................ 48
APPENDIX B: COMPUTATION OF ESTIMATED CSSR ................................................. 54
vii
LIST OF FIGURES
Figure 2.1: GSM Architecture ..................................................................................................... 6
Figure 2.2: GSM Burst Structure ............................................................................................... 10
Figure 2.3: GSM Call Origination (Radio Aspect) ..................................................................... 13
Figure 2.4: GSM Call Termination (Radio Aspect) .................................................................... 13
Figure 2.5: GPRS Network Architecture ................................................................................... 14
Figure 2.6: Correlation between Voice KPIs and End user Experience ...................................... 16
Figure 2.7: Correlation between Data KPIs and End user Experience ........................................ 17
Figure 2.8: Model for calculation of NumCH_ReqSpeech ......................................................... 28
Figure 4.1: Hourly Plot of Measured and Estimated CSSR ........................................................ 34
Figure 4.2: Busy_Hour Plot of Measured and Estimated CSSR ................................................. 35
Figure 4.3: Daily Plot of Measured and Estimated CSSR .......................................................... 36
viii
LIST OF TABLES
Table 2.1: User oriented KPIs…………………………………………………………………...17
Table 2.2: Call Setup Success Rate (CSSR).……………..……………………………………...21
Table 2.3: Call Drop Rate (CDR).…………………………..…………………………………...21
Table 2.4: Handover Success Rate (HSR)……………………..………………………………...22
Table 2.5: TCH Congestion Rate (TCHCR)……………………...……………………………...22
Table 4.1: Regression Analysis of Hourly CSSR Observations ................................................. 37
Table 4.2: Regression Analysis of Busy_Hour CSSR Observations ........................................... 38
Table 4.3: Regression Analysis of Daily CSSR Observations ................................................... 39
Table 4.4: Summary of p-values ............................................................................................... 42
Table 4.5: Summary of F-test statistic.. ..................................................................................... 43
ix
ABBREVIATIONS
3G
Third generation
AGCH
Access Grant CHannel
ARQ
Automatic Repeat Request
AuC
Authentication Center
BCCH
Broadcast Control Channel
BCH
Broadcast Channel
BSC
Base Station Controller
BSS
Base Station System
BTS
Base Transceiver Station
CBCH
Cell Broadcast Channel
CCCH
Common Control Channel
CCH
Control Channel
CDMA
Code Division Multiple Access
CDR
Call Drop Rate
CEPT
Confederation of European Posts and Telecommunications
CS
Circuit Switching
CSSR
Call Setup Success Rate
EDGE
Enhanced Data rates for GPRS Evolution
EFR
Enhanced Full Rate
ETSI
European Telecommunications Standards Institute
FACCH
Fast Associated Control Channel
FCCH
Frequency Correction Channel
FDMA
Frequency Division Multiple Access
FTP
File Transfer Protocol
GGSN
Gateway GPRS Support Node
GMSC
Gateway Mobile Switching Center
GPRS
General Packet Radio Service
GSM
Global System for Mobile communications
HLR
Home Location Register
x
HOSR
HandOver Success Rate
HSCSD
High Speed Circuit Switched Data
HTTP
HyperText Transfer Protocol
ISDN
Integrate Services Digital Network
KPI
Key Performance Indicator
LA
Location Area
MAC
Media Access Control
ME
Mobile Equipment
MO
Mobile Origination
MS
Mobile Station
MSC
Mobile Switching Center
MT
Mobile Terminal
MT
Mobile Termination
NCA
National Communication Authority
NSS
Network and Switching Subsystem
OSI
Open Systems Interconnection
PCH
Paging Channel
PIN
Personal Identity Number
PS
Packet Switching
PSDN
Public Switched Data Network
PSR
Paging Success Rate
PSTN
Public Switched Telephone Network
PUK
PIN Unblocking Key
QoS
Quality of Service
RACH
Random Access Channel
RF
Radio Frequency
RLC
Radio Link Control
SACCH
Slow Associated Control Channel
SCH
Synchronization Channel
SDCCH
Standalone Dedicated Control Channel
SGSN
Serving GPRS Support Node
xi
SIM
Subscriber Identity Module
SMS
Short Message Service
TCH
Traffic Channel
TCH/F
Full rate Traffic Channel
TCH/H
Half rate Traffic Channel
TCP
Transmission Control Protocol
TDMA
Time Division Multiple Access
TE
Terminal Equipment
TRAU
Transcoder/Rate Adapter Unit
TRX
Transceiver
UMTS
Universal Mobile Telecommunications System
VLR
Visitor Location Register
1
CHAPTER ONE
GENERAL INTRODUCTION
In 1895, Guglielmo Marconi [1] opened the way for modern wireless communications by
transmitting the three-dot Morse code for the letter „S‟ over a distance of three kilometers using
electromagnetic waves. From this beginning, wireless communications has developed into a key
element of modern society. From satellite transmission, radio and television broadcasting to the
now ubiquitous mobile telephone, wireless communications has revolutionized the way societies
function.
In the late 1960s and early 1970s [2], the cellular concept was conceived. This was used to
improve system capacity and frequency efficiency. Each cell in a cellular network is equipped
with a base station and a number of radio channels assigned according to the transmission power
constraints and availability of spectrum (bandwidth).
Currently, wireless communication systems are increasingly serving the demands of voice, data
and multimedia users. Previously, these services were considered separate and were optimized
for either voice or data services [3]. High bandwidth internet is becoming available through
Enhanced Data rates for General packet radio service Evolution (EDGE) and the 3rd generation
(3G) mobile systems.
Progression from the first systems to today‟s Universal Mobile Telecommunications System
(UMTS) has resulted in more processing power, more availability and more services due to the
continuing growth of customers and availability of newer technology [4].
2
1.1 Background to the Study
Mobile telephony has become one of the fastest growing and most demanding
telecommunications applications. It represents a continuously increasing percentage of all new
telephone subscriptions around the world. In Ghana, it has assumed a dominant position in the
telecommunications market and become the main driver of growth in the country [5].
New services for mobile phones, like email, web browsing, audio and video streaming demands
a lot from the underlying network. If the network does not deliver what these services demand,
the performance and the user satisfaction will be unsatisfactory. There are different Key
Performance Indicators (KPI) for different services and on different network layers. To identify
these will make it easier to optimize the network and applications. Therefore, it is useful for
companies who specialize in cellular network optimization or even service providers to have the
ability to measure the performance of the network for the purpose of optimizing the network
usage and enhancing customer satisfaction [4].
1.2 Objectives of the Study
Specifically the study sought to;
identify KPIs that affect end user satisfaction in a GSM/GPRS cellular network
identify an aggregative KPI that evaluates network accessibility and service retainability
as perceived by the end user
perform measurements of these top level KPIs using a Nokia Siemens Network Statistics
tool
perform a comparative analysis of the measured aggregative KPI to that predicted or
estimated
3
1.3 Justification of the Study
Network performance management and optimization is necessary to judge network performance
and maintain Quality of Service (QoS) standards. It identifies inconsistencies or limitations in
current overall network design and helps to improve processes. Thus, resulting in optimized
networks and improved quality of service.
1.4 Organisation of the Study
The final report is put into five main chapters. Chapter one provides an introductory background
to the research. Other areas covered include the objectives and justification of the study. Chapter
two aims at giving a greater understanding of relevant technologies and summarizes previous
related works on GSM/GPRS network performance management and optimization. Chapter three
discusses the methodology for the study. Chapter four presents results and analysis. The final
chapter, Chapter five presents conclusion and recommendations.
4
CHAPTER TWO
LITERATURE REVIEW
2.1 Brief History
In 1982, the first step in creating an European mobile communication standard was taken by the
Confederation of European Posts and Telecommunications (CEPT) and the Groupe Spéciale
Mobile was born. Their objective was to design an European standard for mobile communication
to be used all over Europe. In 1989, the responsibility was shifted over to the European
Telecommunications Standards Institute (ETSI) and Global System for Mobile communications
(GSM) was accepted as an international mobile digital telephony standard.
The GSM family of technologies has provided the world with mobile communications since
1991. In over twenty years of development, GSM has been continually enhanced to provide
platforms that deliver an increasingly broad range of mobile services as demand grows. Starting
with plain voice calls, the industry now has a powerful platform capable of supporting mobile
broadband and multimedia services. GSM is now used in 219 countries and territories serving
more than three billion people and providing travellers with access to mobile services wherever
they go [6].
Through the years, the GSM standard has developed when it comes to speech and data, not only
bit rates but also codecs, channel utilization, robustness among others. As the standard evolves,
the data speeds are increasing, quality is improving and the hardware is getting more efficient
[7].
5
2.2 GSM Overview
GSM is an open digital cellular technology used for transmitting mobile voice and data services.
GSM operates in the 900MHz and 1.8GHz bands in Europe and the 1.9GHz and 850MHz bands
in the US. The 850MHz band is also used for GSM and 3G in Australia, Canada and many South
American countries [6]. Each band is divided into one uplink band and one downlink band with
25 MHz bandwidth each. The uplink and downlink bands are subdivided into 124 frequency
channels spaced 200 kHz. These channels are often referred to as carriers or carrier frequencies
[8].
By having harmonised spectrum across most of the globe, GSM‟s international roaming
capability allows users to access the same services when travelling abroad as at home. This gives
consumers seamless and same number connectivity in more than 218 countries.
GSM networks deliver wireless high quality and secure mobile voice and data through a process
called Circuit Switching (CS). Circuit switching is a method of communication that establishes
connections on demand through one or more switching nodes and permits exclusive use of those
connections until they are released. In this networking method, a connection called a circuit is
setup between two devices (e.g. mobile phones) which is used for the whole communication.
Information about the nature of the circuit is maintained by the network (service provider). The
circuit may either be a fixed one that is always present, or it may be a circuit that is created on an
as-needed basis. The digital data is sent as a continuous stream of bits allowing the data to be
heard from the receiving end whilst it is still being sent [5].
6
2.3 GSM Architecture
Figure 2.1 illustrates the GSM architecture. In this architecture, a mobile station (MS)
communicates with a base station system (BSS) through the radio interface. The BSS is
connected to the network and switching subsystem (NSS) by communicating with a mobile
switching center (MSC) using the A interface.
Figure 2.1: GSM Architecture [9]
2.3.1 Mobile Station
The MS consists of two parts: the subscriber identity module (SIM) and the mobile equipment
(ME). In a broader definition, the MS also includes a third part called terminal equipment (TE),
which can be a PDA or PC connected to the ME. In this case, the first two parts (i.e., ME and
SIM) are called the mobile terminal (MT).
7
The SIM although loaned to the subscriber is the property of the service provider. It contains the
subscriber-related information including the personal identity number (PIN) and PIN Unblocking
Key (PUK) codes, a list of abbreviated and customized dialing numbers, short messages received
when the subscriber is not present, names of preferred networks to provide service and so on.
The ME contains the non-customer-related hardware and software specific to the radio interface.
When the SIM is removed from an MS, the remaining ME cannot be used for reaching the
service except for emergency calls. SIMs may be attached to MEs with different characteristics.
At every new connection between MS (SIM) and the network, the characteristic indication of the
ME called classmark is given to the network. This SIM-ME design supports portability as well
as enhancing security. Usually, the ME is the property of the subscriber [9].
2.3.2 Base Station System
The BSS connects the MS and the NSS. It consists of two parts: the base transceiver station
(BTS) and the base station controller (BSC). The BTS contains transmitter, receiver, and
signalling equipment specific to the radio interface in order to contact the MSs. An important
part of the BTS is the transcoder/rate adapter unit (TRAU) that carries out GSM-specific speech
encoding/decoding and rate adaption in data transmission.
The BSC is responsible for the switching functions in the BSS, and is in turn connected to an
MSC in the NSS. It supports radio channel allocation/release and handoff management. The BSC
may connect to several BTSs and maintain cell configuration data of these BTSs. It also
communicates with the BTSs using Integrate Services Digital Network (ISDN) protocols via the
A-bis interface. In GSM BSS design, a BSC may only connect to one BTS in which case they are
8
likely to be collocated. In this scenario, the BSC and the BTS may be integrated without the A-
bis interface [9].
2.3.3 Network and Switching Subsystem
The NSS supports the switching functions, subscriber profiles and mobility management. The
basic switching function in the NSS is performed by the MSC. This interface follows a signalling
protocol used in the telephone network. The MSC also communicates with other network
elements external to GSM utilizing the same signalling protocol.
The current location of an MS is usually maintained by the Home Location Register (HLR) and
Visitor Location Register (VLR) (as described in Section 2.3.3.1 and 2.3.3.2 respectively). When
an MS moves from the home system to a visited system, its location is registered at the VLR of
the visited system. The VLR then informs the MS's HLR of its current location.
The authentication center (AuC) is used in the security data management for the authentication
of subscribers. The AuC may be collocated with the HLR.
An incoming call is routed to an MSC unless the fixed network is able to interrogate the HLR
directly. That MSC is called the gateway MSC (GMSC). An MSC can function as a GMSC by
including appropriate software and HLR interrogation information, and by provisioning interface
and the signalling link to the HLR. The GMSC obtains the location information and routes the
calls to the visited MSC of the subscribers to receive the calls [9].
9
2.3.3.1 Home Location Register
This is a database that contains all the administrative information about each subscriber along
with their last known location. In this way, the GSM network is able to route calls to the relevant
base station for the MS. When a user switches on their phone, the phone registers with the
network and from this it is possible to determine which BTS it communicates with so that
incoming calls can be routed appropriately. Even when the phone is not active (but switched on)
it re-registers periodically to ensure that the network (HLR) is aware of its latest position. There
is one HLR per network although it may be distributed across various sub-centres for operational
reasons [10].
2.3.3.2 Visitor Location Register
This contains selected information from the HLR that enables the selected services for the
individual subscriber to be provided. The VLR can be implemented as a separate entity but it is
commonly realized as an integral part of the MSC rather than a separate entity. In this way
access is made faster and more convenient [10].
2.3.4 Radio Interface
The GSM radio link uses both Frequency Division Multiple Access (FDMA) and Time Division
Multiple Access (TDMA) technologies. The frequency bands for the GSM downlink signal and
the uplink signals are 935-960MHz and 890-915MHz respectively. The frequency band is
divided into 124 pairs of frequency duplex channels with 200 KHz carrier spacing. Note that for
a given distance less power is required to transmit signal over a lower frequency. To save MS
power, uplink frequencies in mobile systems are always the lower band of frequencies.
10
The GSM burst structure is illustrated in Figure 2.2. Every burst contains 148 bits (0.546 msec)
followed by 0.031 msec guard time (8.25 bits). The burst begins with 3 head bits and ends with 3
tail bits all of which are logical zeros.
Figure 2.2: GSM Burst Structure [9]
Two groups of data bits are separated by an equalizer training sequence of 26 bits. Each data
group consists of 57 information bits and one flag that indicate whether the information bits are
for user speech/data or signalling. Depending on the information carried by a time slot, i.e., the
information bits in Figure 2.2, two types of logical channels are defined: the traffic channels
(TCHs) and the control channels (CCHs).
TCHs are intended to carry user information (speech or data). Two kinds of TCHs are defined:
I. Full rate TCH (TCH/F) provides transmission speed of 13 Kb/s for speech or 9.6, 4.8 or
2.4 Kb/s for data. Enhanced Full Rate (EFR) speech coders have been implemented to
improve the speech quality of a TCH/F.
II. Half rate TCH (TCH/H) allows transmission of 5.6 Kb/s speech or 4.8 or 2.4 Kb/s data.
11
The CCHs are intended to carry signalling information. Three types of CCHs are defined in
GSM:
I. Common control channels (CCCHs) include the following channel types:
The paging channel (PCH) is used by the network to page the destination MS in call
termination.
The access grant channel (AGCH) is used by the network to indicate radio link
allocation upon prime access of an MS.
The random access channel (RACH) is used by the MSs for initial access to the
network.
Several MSs may access the same RACH potentially resulting in collisions. The slotted Aloha
protocol is adopted in GSM to resolve access collision. PCH and AGCH are delivered from the
BSS to the MSs by the downlink. RACH utilizes the uplink.
II. Dedicated control channels are supported in GSM for dedicated use by a specific MS.
The standalone dedicated control channel (SDCCH) is used only for signalling and
for short messages.
The slow associated control channel (SACCH) is associated with either a TCH or an
SDCCH. The SACCH is used for non-urgent procedures mainly the transmission of
power and time alignment control information over the downlink and measurement
reports from the MS over the uplink. A TCH is always allocated with a control
channel SACCH to transport both user information and signalling data in parallel.
The fast associated control channel (FACCH) is used for time critical signalling such
as call establishing progress, authentication of subscriber, or handoff. The FACCH
12
makes use of the TCH during a call. Thus, there is a loss of user data because the
FACCH “steals” the bandwidth of the TCH.
The cell broadcast channel (CBCH) only carries the short message service cell
broadcast messages which uses the same time slot as the SDCCH.
The CBCH is used in downlink only. SDCCH, SACCH and FACCH are used in both downlink
and uplink.
III. Broadcast channels (BCHs) are used by the BTS to broadcast information to the MSs in
its coverage area.
The frequency correction channel (FCCH) and the synchronization channel (SCH)
carry information from the BSS to the MS. The information allows the MS to acquire
and stay synchronized with the BSS.
The broadcast control channel (BCCH) provides system information such as access
information for the selected cell and information related to the surrounding cells to
support cell selection and location registration procedures in an MS.
Figure 2.3 shows the radio aspect of mobile call origination which describes how the logical
channels are involved in the call setup procedure. To initiate the call setup, the MS sends a
signalling channel request to the network through RACH. The BSC informs the MS of the
allocated signalling channel (SDCCH) through AGCH. Then the MS sends the call origination
request via SDCCH. The MSC instructs the BSC to allocate a TCH for this call. Finally, both the
MS and the BTS tune to the TCH.
13
Figure 2.3: GSM Call Origination (Radio Aspect) [9]
Figure 2.4 the radio aspect of mobile call termination is shown. In this case the MSC requests the
BSS to page the MS. The BSCs instruct the BTSs in the desired Location Area (LA) to page the
MS by using PCH. When the destination MS receives the paging message, it requests for a
SDCCH. The BTS assigns the SDCCH which is used to setup the call as in the call origination
case.
Figure 2.4: GSM Call Termination (Radio Aspect) [9]
14
2.4 GPRS Overview
GPRS is a packet-based data bearer service for wireless communication services that is delivered
as a network overlay for GSM, Code Division Multiple Access (CDMA) and TDMA networks.
GPRS applies a packet radio principle to transfer user data packets in an efficient way between
GSM mobile stations and external packet data networks. Packet switching is where data is split
into packets that are transmitted separately and then reassembled at the receiving end. GPRS is
based on GSM communication and complements existing services such as circuit switched
cellular phone connections, Short Message Service (SMS) and Bluetooth, a standard for
replacing wired connections between devices with wireless radio connections [11].
GPRS and the packet domain transfers packet in an efficient way and optimized the use of
network and radio resources. Unlike circuit switched communication, resources are not held
when they are not used. To allow the network subsystem to be reused with other radio access
technologies, a strict separation between the radio subsystem and the network subsystem is
maintained. This allows a common packet domain core network for both GSM and UMTS [4].
Figure 2.5 shows the GPRS network architecture.
Figure 2.5: GPRS network Architecture [9]
15
There are two main GPRS nodes in the core network which can be shared with other networks
like UMTS, namely, Serving GPRS Support Node (SGSN) and Gateway GPRS Support Node
(GGSN). SGSN receives and transmits packets between the MSs and their counterparts in the
public switched data network (PSDN). GGSN interworks with the PSDN using connectionless
network protocols such as Internet Protocol and the Open Systems Interconnection (OSI)
connectionless network protocol, or connection oriented protocols such as X.25. SGSN and
GGSN interact with the GSM location databases including the HLRs and the VLRs to track the
location of the MSs. The GPRS data units are routed to the destination MSs based on location
information. Both SGSN and GGSN may be equipped with cache memories containing location
information to speed up the routing procedure.
Like High Speed Circuit Switched Data (HSCSD), GPRS air interface requires a new radio link
protocol to guarantee fast call setup procedure and low bit error rate for data transfer between the
MSs and the BSs. Furthermore, GPRS needs to implement a packet radio media access control
(MAC) for packet switching [9].
16
2.5 Performance Management
Performance management involves capturing and analyzing a set of Key Performance Indicators
(KPI) that can be quantified by using different sources of performance data. Network operators
typically define a set of KPIs and stratified target thresholds in line with marketing and business
priorities [12]. The network operator concentrates on many varied performance indicators
assigning assumed/theoretical priorities while the subscriber judges the network on the basis of a
few basic indicators of performance. It is very important to align the KPI definition according to
what quality and performance mean to the subscriber [13]. Figures 2.6 and 2.7 provide a
correlation between voice and data KPIs, and the subscriber/end user satisfaction (experience)
respectively.
Figure 2.6: Correlation between voice KPIs and end user satisfaction [14]
17
Figure 2.7: Correlation between data KPIs and end user satisfaction [14]
End user satisfaction cannot be represented by a single KPI. A mix of KPIs can be very close to
achieving required targets. Table 2.1 shows some suggested user satisfaction oriented KPIs for
network performance estimation.
Table 2.1: User oriented KPIs [14]
18
Several studies have been conducted on performance evaluation and optimization of cellular
networks in the literature [14 – 17].
In [16], a real GSM radio access network was evaluated. A network audit was performed on the
radio access part of the GSM network using an audit methodology algorithm or flow chart. The
algorithm included many aspects of the network such as performance, neighbor, parameter,
frequency and competitive benchmark audits. The audit was mainly a comparative process which
required the initial establishment of baseline KPIs. These KPIs were derived from design
guidelines, service requirements, customer expectations and market benchmarks. Two
assessment parameters (network accessibility and service retainability) for evaluating the
performance on the network were employed.
Network Accessibility
Accessibility is the ability of a service to be obtained within specified tolerances and
other given conditions when requested by the user.
(2.1)
Listed below are the KPIs connected to accessibility.
I. Paging Success Rate (PSR)
The paging success rate is a measure of the percentage of how many paging attempts that
have been answered either as a result of the first or the second repeated call.
(2.2)
II. SDCCH Access Success Rate
SDCCH access success rate is a percentage of all SDCCH accesses received in the BSC.
19
III. SDCCH Drop Rate
The SDCCH drop rate compares the total number of RF losses (while using an SDCCH)
as a percentage of the total number of call attempts for SDCCH channels. This statistic is
intended to give an indication of how good the cell/system is at preserving calls.
(2.3)
IV. Call Setup Success Rate (CSSR)
This measures successful TCH assignments of total number of TCH assignment attempts.
(2.4)
(2.5)
V. Call Setup TCH Congestion Rate
The Call Setup TCH Congestion Rate statistic provides the percentage of attempts to
allocate a TCH call setup that were blocked in a cell.
(2.6)
20
Service Retainability
Service retainability is the ability of a service once obtained to continue to be provided
under given conditions for a requested duration. In other words:
(2.7)
Listed below are the KPIs connected to retainability.
I. Call Drop Rate (CDR)
This KPI gives the rate of drop calls by percentage of TCH drops after TCH assignment
is completed.
(2.8)
DR is directed retry
II. Handover Success Rate (HOSR)
The handover success rate shows the percentage of successful handovers of all handover
attempts. A handover attempt is when a handover command is sent to the mobile.
(2.9)
Haider et al [18], established a real GSM RF network performance evaluation on the basis of
four major KPIs i.e., Call Setup Success Rate (CSSR), Call Drop Rate (CDR), Handover Success
Rate (HSR) and Radio traffic channel (TCH) congestion rate. These KPIs were summarized as
shown in Tables 2.2 to 2.5. The study described simple procedures for cellular network
performance estimation. It was analytically proven that optimizing an existing cellular network
using different methodologies and fine parameter tuning to offer remarkable QoS to the end
users is possible.
21
Table 2.2: Call Setup Success Rate (CSSR) [18]
Indicator CSSR
Definition Rate of call attempts until TCH successful assignment
Formula
Result *100
Condition
Applied
Where counter CT01 counts SDCCH channels successfully seized for Call
termination & CT02 counts SDCCH channels successfully seized for Call
origination. CT03 counts SDCCH seizure requests.
Table 2.3: Call Drop Rate (CDR) [18]
Indicator CDR
Definition Rate of calls not completed successfully.
Formula
Result *100
Condition
Applied
Where CT04 counts TCH drops due to radio interface problems & CT05 counts
TCH drops due to BSS problems. CT06 counts numbers of TCH successfully
seized/assigned.
22
Table 2.4: Handover Success Rate (HSR) [18]
Indicator HSR
Definition Rate of successful handovers (intercell + intracell).
Formula
Result
*100
Condition
Applied
Where CT07 counts number of incoming successful handovers & CT08 counts
number of outgoing successful handovers. CT09 counts number of outgoing
handover requests while CT10 counts number of incoming handover requests.
Table 2.5: TCH Congestion Rate (TCHCR) [18]
Indicator TCHCR
Definition Rate of blocked calls due to resource unavailability.
Formula
Result *100
Condition
Applied
Where CT11 counts number of assignment failures when there is no TCH
available & CT12 counts number of normal assignment requests for TCH
establishment.
23
In [19], a summary of measurements and trials that were carried out and evaluations of the
performance of GSM and GPRS were presented. Their study revealed that limitations existed in
the GSM system with regards to accommodating extreme offered traffic. Also, the GSM system
could not predict the rapidly increased traffic in many cases and it could definitely not adapt
even by reconfiguring system parameters. In their opinion, GSM was not yet optimized and
GPRS, on the other hand was still immature and several issues needed to be considered.
Orstad and Reizer [4], performed practical end-to-end performance tests in cellular networks
using an end-to-end test agent, TWSE2E. Their objective was to identify what affects end user
performance. Special attention was given to the high latency of the wireless links and the delay
introduced with the radio access bearer establishment. They concluded that the 3G cellular
network UMTS outperformed EDGE with respect to commonly used services like HTTP/WEB
and FTP. It was also discovered that while TCP throughput was good when transferring large
files over FTP, the latency of the wireless link made the HTTP performance bad compared to
potential TCP throughput.
Meyer in [20] investigated performance issues of TCP when operated over GPRS. Several traces
for TCP and RLC were discussed to highlight the characteristics if GPRS is applied as access for
bulk data download. Download time measurements showed that GPRS could operate effectively
over a wide range of channel conditions. Further results revealed that GPRS provided a
sufficiently fast working ARQ mechanism based on the RLC protocol which allowed typically
several retransmissions before TCP timed out. He concluded that the RLC ARQ scheme is
appropriately designed to ensure that TCP observes just packet delays rather than losses. If
additionally these delays could be bounded in such a way that no TCP time-out occurs, TCP
would not notice packet errors on the wireless link at all.
24
Adolfsson [21], by simulation identified possible problems when running TCP in a GPRS
environment. Some of these problems he noted were mostly caused by the TBFs which are setup
between the MS and the BSC whenever data should be sent. The buffers and the slow link
between mobile station and base station may cause other problems which are not fatal but may
decrease TCP performance or make TCP less able to respond to loss of data. TCP features that
are especially important for good performance were also identified. These features he noted
mostly dealt with fast recovery from data loss and how to avoid data loss because of small
buffers in the GPRS system. He further suggested some improvements to TCP performance
such as responding to medium access requests from the MS quickly, using Explicit Congestion
Notification (ECN) to avoid unnecessary drops from the PCU or SGSN queue, implementing
new standards for delayed TBF release which would improve connection setup and maybe
improve the acknowledgement clustering situation.
In [22], an approximation method is used for evaluating the GPRS performance of single-slot
service in the variable radio resource. Voice services are independent of GPRS and because
GPRS is mainly designed to transmit intermittent and burst data, the service time of GPRS is
rather smaller than that of voice services. As an approximation, the decomposition technique was
used to analyze the GPRS performance. The essence of this technique was to use the voice
services probability distribution to describe the interaction of voice services to GPRS. Thus, the
GPRS performance in the dynamically variable resource was obtained by combining this
distribution with the performance in a fixed resource.
By the comparison of numerical results and simulated results, it was shown that the method
could be used for evaluating GPRS performance when the average service time of circuit
switched services is much longer than that of GPRS. The simulations showed that the
25
interruption probability of GPRS calls due to releasing its channel to the demand of circuit
switched services depended on the average message size more strongly than on the traffic load.
The multi-slot services caused higher blocking probability and longer delay to the network than
the single-slot service. These effects they observed could be reduced by implementing a GPRS
resource allocation scheme with flexible multi-slot services. In this scheme, when the available
network resource cannot provide a call with its required transmission rate, the network negotiates
with the user (GPRS call or circuit switched service) and agrees on a transmission rate which the
network can provide.
Kollár [17], gave a definition of a real Call Setup Success Rate (CSSR) and the possibility of its
implementation using current GSM technologies. It was concluded that more complex
formulation which utilized the Immediate Assignment Success rate, TCH Assignment Success
Rate and SDCCH Success rate must be used for measuring CSSR. He further stated that this
formulation was the best approach despite a higher effort on the processor part of the equipment
where the CSSR is to be calculated. He noted that the formulation did not cover the case when
the Direct TCH Assignment feature is enabled.
The aggregative KPI ability that evaluates network accessibility and service retainabilty as
perceived by the end user is a Call Setup Success Rate. This consists of three main voice call
KPIs [17]:
Successful Immediate assignment procedure (the result is occupation of SDCCH or
FACCH in case of Direct TCH assignment)
Successful authentication and ciphering on SDCCH or FACCH (these procedures can be
excluded in case of Direct TCH assignment)
Successful TCH assignment
26
Therefore, equation 2.4:
can also be written as;
(2.10)
where NumTCHAssig represents the number of successfully assigned TCH (number of
ASSIGNMENT Complete messages) and NumCH_ReqSpeech represents the number of
CHANNEL REQUEST messages but related only to request for a mobile originated (MO) or
mobile terminated (MT) call. The other procedures, which can be completed with an SDCCH
like SMS – MT, SMS – MO, location updating etc. are not counted because they do not represent
the request for the speech call. The practical implementation of the equation 2.10 is problematic
because up to now it is not possible to distinguish between the requests for the speech call and
other calls [17].
One possibility of solving this problem is using the simplified formula given in equation 2.11:
= (2.11)
where NumCH_Req represents total number of CHANNEL REQUEST messages and
NumTCHAssig represents the number of TCH assignments (number of ASSIGNMENT Complete
messages). Given that;
(2.12)
where NumCH_ReqNonSpeech is the number of CHANNEL REQUEST messages not used for
MT or MO speech call. Then equation 2.11 can be modified to give equation 2.13:
(2.13)
Under the condition that equation 2.13 can be modified using
binomial series as follows:
27
(2.14)
The absolute error in measurement of Call Setup Success Rate using equation 2.11 is then
evaluated as given in equation 2.15;
(2.15)
In the case that NumCH_ReqNonSpeech is equal to zero, equation 2.11 provides exactly the call
setup success rate. However, in practice the ratio; is within the ranges of
tenths of percent which can lead to a big systematic error. Therefore, the mobile operators break
away from the use of equation 2.11. In principle equation 2.11 can be used for the calculation of
call setup success rate only in regions where the
The second possibility is to replace the denominator (NumCH_ReqSpeech) in equation 2.10 with
NumTCHAttempt where NumTCHAttempt represents the number of ASSIGNMENT REQUEST
messages. But in this case the result of the calculation will be TCH Assignment Success rate
which is something different from Call Setup Success Rate [17]. Even some of the operators
have separate KPIs for Call Setup Success Rate and TCH Assignment Success Rate.
The best approach promises the indirect calculation of NumCH_ReqSpeech according to the
model given in Figure 2.8.
28
Figure 2.8: Model for calculation of NumCH_ReqSpeech [17]
From Figure 2.8, ImmAssSuccRate represents Immediate Assignment Success Rate given by the
relationship:
(2.16)
where NumEstInd represents the number of ESTABLISH INDICATION messages. In other
words, the Immediate Assignment Success Rate represents the total number of requests for
channels that were successful during the immediate assignment procedure.
SDCCHSuccRate represents SDCCH Success Rate and is given by:
(2.17)
where SDCCH_Drop_Rate is SDCCH Drop rate and provides the total number of SDCCH
dropped during the procedures (authentication, ciphering etc.) performed on SDCCH.
From the model in Figure 2.8,
(2.18)
NumCH_ReqSpeech in equation 2.18 can be written as;
(2.19)
Substituting equation 2.19 into equation 2.10 gives;
29
(2.20)
which can be rewritten as given in equation 2.21;
(2.21)
Equation 2.21 is currently the best approach used in measuring the Call Setup Success Rate
(CSSR). A disadvantage could be higher effort on BSC or the equipment where the CSSR is to
be calculated as three KPIs (or six partial measurements) are used. It provides exactly the CSSR
in the case where Direct TCH Assignment feature is disabled [17].
30
CHAPTER THREE
METHODOLOGY
3.1 Research Statement
The study aims to present an insight into network performance evaluation of a GSM/GPRS
cellular network by conducting some measurements. One of the most important KPIs used by all
mobile operators is the Call Setup Success Rate (CSSR). The work carries out a comparative
analysis of the measured CSSR to that estimated using a CSSR formulation where the direct
Traffic Channel (TCH) Assignment feature is disabled.
3.2 Data Collection
The performance measurements were conducted on ONATELs GSM/GPRS cellular network
using a Nokia Siemens Network Statistics tool to define top level KPIs which describe the
success/failure rates of the most important events such as service blocking, service dropping and
handovers at the BTS level.
3.3 Sample size and data processing
The data collection was over a four (4) week period and categorized into the following
observation time intervals [23]:
Hour: Hourly statistics give a detailed picture of the network performance and are useful
to help spot temporary problems and identify trends.
Peak or busy Hour: Peak hour statistics are of great significance because they correspond
to the time of heavy utilization of network resources. In a way, they provide the “worst-
case” scenario.
31
Day: Daily statistics are introduced to provide a way of averaging temporary fluctuations
of hourly data. Problems can be identified and corrective actions triggered with more
confidence.
Data processing was done using Microsoft Excel (see appendix).
32
CHAPTER FOUR
RESULTS AND ANALYSIS
In this section results that have come up through the comparison of operational data to that
estimated using the CSSR formulation as given in equation 2.21:
is presented.
The logical channels that are primarily used in today‟s mainly, voice traffic in cellular networks
are the Traffic Channel (TCH) and Standalone Dedicated Control Channel (SDCCH) often
referred to as “signalling channels”. Though many other channels exist, these two (especially the
SDCCH) are the most important resources where the system relies on in order to accommodate
the subscribers needs [24]. A new call cannot be initiated if SDCCH channels are not available
and the same happens when SDCCHs are available but all TCHs are blocked. Thus one can say
that blocking of these channels is a main performance indicator for an operational GSM/GPRS
cellular network that may lead to severe bottlenecks if the phenomenon persists.
The SDCCH and TCH Success Rates KPIs provide an understanding of when and where
congestion appears since these channels are the most vulnerable and directly affects the quality
of service offered to the subscribers.
In Figures 4.1 to 4.3 and Tables 4.1 to 4.3, the graphical representation and regression analysis of
the measured and estimated CSSR for the various observation time intervals are shown
respectively. The top level KPI measurements including the measured CSSR and the
computation of the estimated CSSR are shown in appendices A and B respectively. In appendix
B the SDCCHdropRate and TCHdropRate are substracted from 100% to obtain the
33
SDCCHsuccRate and TCHsuccRate respectively. Also, the IMM_ASSGNsuccRate is obtained
by subtracting the outcome of equation 4.1 from 100%.
(4.1)
34
4.1 Presentation of Results
4.1.1 Graphical Representation of Measured and Estimated CSSR
Figure 4.1: Hourly Plot of Measured and Estimated CSSR
0.0010.0020.0030.0040.00
50.00
60.00
70.00
80.00
90.00
100.00
188
17
526
234
943
652
361
069
778
4
871
958
10
45
1132
1219
1306
13
93
1480
15
67
1654
1741
1828
1915
2002
20
89
2176
2263
2350
24
37
2524
2611
2698
2785
2872
CSS
R (%
)
Hourly Observations (hrs)
Measured
Estimated
35
Figure 4.2: Busy_Hour Plot of Measured and Estimated CSSR
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
1 5 9
13 17
21 25
29 33 37 41 45 49 53 57 61 65 69 73 77 81 85
89 93 97
101
105
109
113
117
12
1
CSS
R (%
)
Busy_Hour Observations (hrs)
Measured
Estimated
36
Figure 4.3: Daily Plot of Measured and Estimated CSSR
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
1 5 9
13 17 21
25 29
33 37 41
45 49 53 57 61
65 69 73 77 81 85 89 93 97
10
1
105
109
11
3
117
121
CSS
R (%
)
Daily Observations (days)
Measured
Estimated
37
4.1.2 Regression Analysis of Estimated CSSR
Table 4.1: Regression Analysis of Hourly CSSR Observations
HOURLY SUMMARY OUTPUT
Regression Statistics Multiple R 0.999856442 R Square 0.999712905 Adjusted R Square 0.999712612 Standard Error 0.066270816 Observations 2943
ANOVA
df SS MS F Significance
F Regression 3 44946.25236 14982.08 3411360 0 Residual 2939 12.90756216 0.004392
Total 2942 44959.15993
Coefficients Standard
Error t Stat P-value Lower 95% Upper 95%
Lower 95.0%
Upper 95.0%
Intercept -192.5234808 0.116536446 -1652.05 0 -192.751982 -192.295 -192.75198 -192.29498
IMM_ASSGN_Success Rate 0.975515392 0.000788655 1236.936 0 0.97396902 0.977062 0.97396902 0.97706176
SDCCH Success Rate 0.990799239 0.000403807 2453.646 0 0.99000747 0.991591 0.99000747 0.99159101
TCH Assignment Success Rate 0.95866157 0.000870567 1101.192 0 0.95695459 0.960369 0.95695459 0.96036855
38
Table 4.2: Regression Analysis of Busy_Hour CSSR Observations
BUSY_HOUR SUMMARY OUTPUT
Regression Statistics Multiple R 0.999787297 R Square 0.999574639 Adjusted R Square 0.999563916 Standard Error 0.070670991 Observations 123
ANOVA df SS MS F Significance F
Regression 3 1396.649275 465.5498 93214.56 2.2563E-200 Residual 119 0.594332295 0.004994
Total 122 1397.243608
Coefficients Standard
Error t Stat P-value Lower 95% Upper 95%
Lower 95.0%
Upper 95.0%
Intercept -188.5801282 0.878199492 -214.735 6.2E-156 -190.3190509 -186.841 -190.319051 -186.841206
IMM_ASSGN_Success Rate 0.982417884 0.007158682 137.2345 6.9E-133 0.968242979 0.996593 0.96824298 0.99659279
SDCCH Success Rate 0.978913216 0.002981544 328.3243 7.8E-178 0.973009461 0.984817 0.97300946 0.98481697
TCH Assignment Success Rate 0.923538758 0.006036179 153.0006 1.8E-138 0.911586522 0.935491 0.91158652 0.93549099
39
Table 4.3: Regression Analysis of Daily CSSR Observations
DAILY SUMMARY OUTPUT
Regression Statistics Multiple R 0.999921407 R Square 0.999842819 Adjusted R Square 0.999838857 Standard Error 0.037401178 Observations 123
ANOVA Df SS MS F Significance F
Regression 3 1058.889104 352.963 252324.1 4.2475E-226 Residual 119 0.166462922 0.001399
Total 122 1059.055567
Coefficients Standard
Error t Stat P-value Lower 95% Upper 95%
Lower 95.0% Upper 95.0%
Intercept -189.7674752 0.495432383 -383.034 8.6E-186 -190.7484808 -188.786 -190.74848 -188.78647
IMM_ASSGN_Success Rate 0.973002503 0.003155275 308.3733 1.3E-174 0.966754743 0.97925 0.96675474 0.979250263
SDCCH Success Rate 0.983454262 0.001760516 558.6171 2.8E-205 0.979968265 0.98694 0.97996826 0.986940259
TCH Assignment Success Rate 0.940587294 0.004327688 217.3418 1.5E-156 0.932018041 0.949157 0.93201804 0.949156548
40
4.2 Interpretation of Results
4.2.1 Interpretation of Graphical Representations
Halonen et al [25], stated that a well-performing network should have both Call setup success
rate and Handover success rate above 95%. Using this as benchmark and referring to the
graphical representations in Figures 4.1 to 4.3 for the three observation time intervals, it can be
said that averagely, the results of the measured CSSR are within acceptable levels though a
number of sharp drops in CSSR recordings exist. The estimated CSSR, on the other hand, shows
significant improvements in areas where sharp drops in CSSR values were recorded for the
measured CSSR.
A fall in the CSSR below the 95% benchmark were as a result of an increased demand for
establishment resources whiles resources were fixed and limited, i.e., blocking increases
proportionally to demand. Approaching high congestion situations and maximum handling of
traffic, services‟ requests begin not to be served, thus decreasing TCH Blocking and Handover
Failure rates. This results in an increased TCH success rate because the network runs out of call
establishment resources (SDCCH mostly). The real call traffic is reduced (as users terminate
calls) and TCHs become available. It is worth noting that in cellular networks, TCHs outnumbers
SDCCHs in every single Transceiver (TRX) [26].
If conditions worsen more, even deeper beyond the high congestion threshold, then the user
cannot even access the network, i.e., Radom Access Channel (RACH) Blocking starts to appear
and the user cannot send a Channel Request message at all because the network cannot
„understand‟ these attempts [22].
A significant observation made from the analysis was the possibility of below standard
performance recordings for CSSR being caused by other reasons for which measurements were
41
not available. The most common reasons for such failures are the inadequate radio signal
propagation conditions (problems in radio channel in the air interface). The failures in radio
channel are usually due to bad signal quality, i.e., the transmitted data includes too many bit
errors. Also, the radio signal quality is mostly affected by two components. Firstly, the
propagation environment causes attenuation to the transmitted radio signal due to path loss,
shadow fading and multipath fading. Secondly, the radio signal may be attenuated by the other
radio signals originating from other BTSs having a TRX on the same physical frequency
(interference).
4.2.2 Interpretation of Regression Analysis of Estimated CSSR
Evaluation of R square value
An important aspect of a statistical procedure that drives a model from empirical data is
to indicate or show how well the model predicts results. A widely used measure of the
predictive efficacy of a model is its coefficient of determination (R square). If there is a
perfect relation between the dependent and independent variables, R square is 1. Where
there is no relationship between the dependent and independent variables, R square is 0.
From Tables 4.1 to 4.3, the regression statistics show an R square value of about 0.9997,
0.9996 and 0.9998 for hourly, busy hour and daily observations respectively. This means
that a percentage of about 99.97%, 99.96% and 99.98% (representing hourly, busy hour
and daily observations respectively) of the variance in CSSR is expressed in terms of
IMM_ASSGN_Success Rate, SDCCH Success Rate and TCH Assignment Success Rate
indicating a high predictive ability.
42
Test of Statistical Significance (using the -value approach, )
The -value shows how likely it is that the coefficient for an independent variable
emerged by chance and does not describe a real relationship. For instance, a -value of
0.05 means that there is a 5% chance that the relationship emerged randomly, and a 95%
chance that the relationship is real.
Condition I: IF THEN reject the null hypothesis.
Condition II: IF THEN accept the null hypothesis.
Table 4.4 shows a summary of the -values drawn from the regression analysis for each
independent variable, IMM_ASSGN_Success Rate, SDCCH Success Rate and TCH
Assignment Success Rate for the various observation time intervals. -values of less than
for all three independent variables for the various observation time intervals were
recorded. Hence, the null hypothesis is rejected and it can be said that there is a
significant relationship between the estimated CSSR and the independent variables. In
other words, IMM_ASSGN_Success Rate, SDCCH Success Rate and TCH Assignment
Success Rate have a positive and significant effect on the estimated CSSR. -values of
zero (0) as recorded in the case of the hourly observations further lay emphasis on the
fact that the independent variables impact significantly on the estimated CSSR.
Table 4.4: Summary of -values
IMM_ASSGN_Success
Rate
SDCCH Success
Rate
TCH Assignment
Success Rate
Hourly 0 0 0
Busy Hour 6.9E-133
7.8E-178
1.8E-138
Daily 1.3E-174
2.8E-205
1.5E-156
43
Overall Test of Significance of Regression Parameters (using the F-test)
F-test shows the significance level for the model as a whole. It measures the likelihood
that the model as a whole describes a relationship that emerged at random, rather than a
real relationship. The lower the significance -value, the greater the chance that the
relationships in the model are real.
From Table 4.5, a summary of the significance -values drawn from the regression
analysis is shown. Significantly low -values were recorded for the three observation
time intervals. Therefore, it can be said that there is a very great chance that the
relationships in the model are real and the conclusion drawn that the regression
parameters are jointly statistically significant.
Table 4.5: Summary of F-test statistic
significance -value
Hourly 0
Busy Hour 2.2563E-200
Daily 4.2475E-226
44
CHAPTER FIVE
CONCLUSION AND RECOMMENDATION
Operator competency in managing performance and optimizing QoS is not easily taught, it is
developed, rather, mainly through trial and error [23]. It is important for mobile network
operators to ensure stability and efficiency to deliver a consistent, reliable and high-quality end
user (subscriber) satisfaction. For network operators the end user perceived QoS is one of the
major forces behind subscriber growth. Thus, it is very important for operators to align their KPI
definitions according to what quality and performance means to the subscriber [13-14].
The CSSR is one of the most important KPIs used by all mobile operators. However, there is no
standard measurement possible for this parameter [17]. In this study, a CSSR formulation for
analyzing GSM network performance in the case where the direct TCH Assignment feature is
disabled as presented in [17] was evaluated. Significantly, high R square values of close to 1
were recorded from the regression analysis. This means that knowing the regressors or
independent variables (IMM_ASSGN_Success Rate, SDCCH Success Rate and TCH
Assignment Success Rate) helps predict the dependent variable (in this case estimated CSSR)
very well. It also means that close to 100% of the estimated CSSR around its mean is explained
by the regressors. This indeed points to the fact that the CSSR formulation is efficient.
The competitive arena for mobile network operators has changed dramatically over the last few
years. It is strongly recommended that the National Communication Authority (NCA) extends
the implementation of the CSSR formulation to CSSR measurements. This would guarantee
quality in the measurement of CSSR by mobile network operators and ensure higher quality of
service delivery to the subscriber.
45
REFERENCES
[1] J.S.Gans, S.P. King and J.Wright.(1999). “Wireless Communication.” Handbook of
Telecommunications Economics. [Online]. 2, pp.2.
Available: http://profile.nus.edu.sg/fass/ecsjkdw/WirelessCommunications_Final.pdf
[March 22, 2009].
[2] M. Ghaderi and R. Boutaba. (2006). “ Call Admission Control in Mobile Cellular Networks:
A Comprehensive Survey.” [Online]. pp.1-6.
Available: bcr2.uwaterloo.ca/~rboutaba/Papers/Journals/WCMC-04.pdf [July 20, 2008].
[3] D.E. Vannucci and P.J. Chitamu. “Efficient Radio Resource Allocation in a GSM and GPRS
cellular network.” Systemics, cybernetics and informatics. [Online]. 2 (5).
Available: www.iiisci.org/journal/CV$/sci/pdfs/P994083.pdf [March 22, 2009].
[4] B. M. Orstad and E. Reizer. “End-to-end key performance indicators in cellular networks.”
Masters thesis, Agder University College, Norway, 2006.
[5] O.K. Darkwa. (2007). “Mobile Telephone Markets in Ghana: Status and Outlook.”
Internet: conf2007.cict.dtu.dk/papers/ [May 29, 2008].
[6] “GSM Technology.” Internet: http://www.gsmworld.com [Nov. 24, 2010].
[7] “2.5G.” Internet: www.mobilein.com [Nov. 30, 2010].
[8] “GSM” Internet: http://www.nzconnections.net.nz/GSM [Nov. 24, 2010].
[9] J. Y. – B. Lin. “GSM System Overview.” Internet : liny.csie.nctu.edu.tw/ch09A4.pdf
[Nov. 24, 2010].
[10] “GSM Network Overview.” Internet: www.radio-electronics.com [Nov. 24, 2010].
[11] “General Packet Radio Service: A white paper by Usha Communications Technology.”
Internet: www.mobilein.com/GPRS.pdf [Nov. 24, 2010].
46
[12] D. Michel. “GPRS performance management and optimization.” [Online].
Available: www.scribd.com/doc/50371011/GPRS-Performance-Management-Optimization
[April 19, 2011].
[13] S. Raina.(2002) “Optimization and Performance measurement of GSM/GPRS networks.”
[Online] Available: http://www.tmforum.org [April 19, 2011].
[14] “Performance Improvement Solution.”
Internet: www.triangulum-telecom.com/src/Performance_Improvement_Solution.pdf
[April 23, 2011].
[15] “KPI Formula.” Internet: www.scribd.com/doc/32692555/KPI-Formula [April 23, 2011].
[16] M. Ali, A. Shehzad and M. A. Akram. “Radio Access Network Audit & Optimization in
GSM.” International Journal of Engineering & Technology [Online]. 10 (1).
Available: www.ijens.org/108401-2727%20IJET-IJENS.pdf [April 19, 2011].
[17] M. Kollár. (2008). “Evaluation of Real Call Set up Success Rate in GSM.” Acta
Electrotechnia et Informatica [Online]. 8(3).
Available:www.scribd.com/doc/49093385/EVALUATION-OF-REAL-CALL-SET-UP-
SUCCESS-RATE-IN-GSM [April 19, 2011].
[18] B. Haider, M. Zafrullah and M.K Islam. (2009). “Radio frequency Optimization & QoS
Evaluation in Operational GSM Network.” Proceedings of the World Congress on Engineering
and Computer Science [Online]. 1. Available: www.iaeng.org/publication/WCECS2009/
WCECS2009_pp393-398.pdf [April 19, 2011].
[19] S. Kyriazakos, N. Papaoulakis, D. Nikitopoulos, E. Gkroustiotis, C. Kechagias, C.
Karambalis and G. Karetsos. (2002). “A Comprehensive Study and Performance Evaluation of
Operational GSM and GPRS Systems under Varying Traffic Conditions.” [Online]
Available: newton.ee.auth.gr/ist2002/papers/SessionT4/2502089.pdf [April 19, 2011].
47
[20] M. Meyer. (1999). “TCP performance over GPRS.” [Online].
Available: www.cs.helsink.fi/u/gurtov/reiner/wcnc99.pdf [April 19, 2011].
[21] K. Adolfsson. “TCP performance in an EGPRS system.” Masters thesis, Linkӧ pings
University, Sweden, 2003.
[22] S. Ni and S. –G. Hggman. (2002). “GPRS performance estimation in GSM circuit switched
services and GPRS shared resource systems.” [Online].
Available: citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.3.7388. [Nov. 24, 2010].
[23] M. Pipikakis. “Evaluating and Improving the Quality of Service of Second-Generation
Cellular Systems.” Bechtel Telecommunications Technical Journal. 2(2), September, 2004.
[24] M. Rahnema. (1993). “Overview of the GSM System and Protocol Architecture.” IEEE
Communications Magazine. [Online].
Available: www.tfe.umu.se/courses,systemteknik/Telesystem/Rahnema.pdf [April 19, 2011].
[25] T. Halonen, J. Romero and J. Melero. (2003). GSM, GPRS and EDGE Performance :
Evolution Towards 3G/UTMS. (2nd
edition). [Online]. Available: books.google.com
[April 19, 2011].
[26] B. H. Walke. (1999). Mobile Radio Networks: Networking and Protocols. (2nd
edition).
[Online].
Available: ebookee.org/Mobile-Radio-Networks-Networking-and-Protocols_343553.html
[April 19, 2011].
48
Appendix A: Top Level KPIs
49
50
51
52
53
54
Appendix B: Computation of Estimated CSSR
55
56